AI Paper Atlas

Research map / Label-efficient and robust learning / Domain adaptation and generalization

A Comprehensive Survey on Source-Free Domain Adaptation

Jingjing Li, Zhiqi Yu, Zhekai Du, Lei Zhu, Heng Tao Shen

TPAMI 2024 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Source-free domain adaptation · 323 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
upgrade SHOT to SHOT++ (pseudo-labeling plus entropy minimization, refined)SHOT · Office-31Avg. accuracy (%): 88.6 -> 89.2Improved
add contrastive learning to pseudo-labeling (HCL+SHOT)SHOT · Office-31Avg. accuracy (%): 88.6 -> 89.8Improved
add virtual domain generation and pseudo-labeling on top of SHOT++ (Kundu et al.)SHOT++ · Office-31Avg. accuracy (%): 89.2 -> 90.7Improved
add BMD class-balanced pseudo-label refinement to SHOT++SHOT++ · Office-31Avg. accuracy (%): 89.2 -> 90.1Improved
add self-attention (transformer) to pseudo-labeling (TransDA)pseudo-labeling SFDA · Office-31Avg. accuracy (%): 88.6 -> 90.7Improved
use source-free adaptation instead of target-supervised oraclebest SFDA method · Office-31Avg. accuracy (%): 3.6% lower than target supervisionHurt
use source-free adaptation instead of target-supervised oracleTransDA · Office-HomeAvg. accuracy (%): 2.7% lower than target supervisionHurt
use source-free adaptation instead of target-supervised oracleBMD + SHOT++ · VisDA-CAvg. accuracy (%): 0.9% lower than target supervisionHurt
adapt with SFDA instead of source-only modelSHOT · Office-31Avg. accuracy (%): 76.1 -> 88.6Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy